Optics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Tuesday 1 September
Magnetism, a fundamental property of solids arising from the collective alignment of electronic spins of magnetic atoms, has recently been expanded by the discovery of altermagnetism, a symmetry-driven phase distinct from both ferromagnetism and antiferromagnetism. Initial efforts to emulate altermagnetism in photonic systems are also being pursued, primarily through the magneto-optic response of materials; however, their intrinsic frequency limitations pose challenges for extending photonic altermagnetism into the optical regime, thereby restricting their material choices and optical applications. Here, we establish a microscopic framework of photonic magnetism through group theory symmetry analysis, defining photonic counterparts of electron spin and magnetic atoms. This approach enables the realization of photonic ferromagnetic, antiferromagnetic, and altermagnetic phases within structured optical lattices constructed by purely dielectric, nonmagnetic materials, which not only broadens the photonic magnetism concept, but also lifts the material and frequency limitations. Analytical models, numerical simulations, and experimental measurements reveal spin-momentum locking directly corresponding to their electronic analogs, showing d-, g-, and i-wave circular dichroism (CD) splitting bands at telecom wavelengths (around 1550 nm). These results demonstrate that altermagnetic order, and magnetism more broadly, can be reconstructed in photonic systems through symmetry and mode coupling alone, opening a new route towards spin-dependent light control and bosonic analogs of quantum magnetic phenomena in dielectric photonic crystals.
Fluid Dynamics
Scaling of boundary-layer disturbances exposed to free-stream turbulence
We present theoretical results related to the experimental findings of Matsubara and Alfredsson (2001) on the scaling of the energy spectra of the Klebanoff modes, i.e. streamwise-elongated vortical disturbances generated by free-stream turbulence in a flat-plate transitional boundary layer. The scaling is explained by a model that describes the streamwise evolution of the streamwise and spanwise energy spectra. The theoretical framework is based on the quasi-steady asymptotic solution of the boundary-region equations, on an axial-symmetric model of the free-stream spectrum, and on the spectral response of the boundary layer to the external perturbations.
Read the full abstract
Analysis of PDEs
Boundary asymptotics for two-dimensional vectorial Allen-Cahn systems
We study asymptotic properties of critical points of the two-dimensional vectorial Allen-Cahn energy with finitely many non-degenerate wells, subject to a homogeneous Neumann boundary condition. For any sequence with uniformly bounded energy, we prove that the limiting full and potential measures are supported on a closed countably 1-rectifiable set up to the boundary and satisfy the discrepancy relations. The potential measure defines a free-boundary stationary rectifiable varifold. Boundary mass may occur: weights are constant on regular boundary arcs, finite-type junctions obey projected balance, and the sole non-boundary branch meets the boundary orthogonally. This provides a Neumann-boundary extension of Bethuel's planar interior theory. The key idea is to read the boundary geometry from the limiting stress-energy tensor, which identifies the potential-energy measure as the stationary interfacial measure even in the presence of boundary concentration.
Read the full abstract
Artificial Intelligence
Escaping Reasoning Basin Collapse with History-Biased Search
Inference-time search with large language models (LLMs) often concentrates on a small set of structurally or semantically similar trajectories, leaving alternative reasoning strategies underexplored—a failure mode we call reasoning basin collapse. We introduce BASIN, a training-free, history-biased search method that groups reasoning states into basins and accumulates a revisit penalty on repeatedly selected basins, reallocating a fixed inference budget toward underexplored reasoning strategies. Under matched inference budgets, BASIN improves over Tree of Thoughts (ToT) by up to $+22$pp on Game of 24 and $+6.7$pp on MuSR. Because indiscriminate diversification can over-explore once search has found a promising basin, we further introduce QA-BASIN, a quality-aware variant that weakens the revisit penalty for high-quality basins and yields more robust gains. To characterize when basin-aware search helps, we introduce the redundancy gap $\Delta$, which measures the difference in search concentration between correct and incorrect predictions: standard ToT often operates near $\Delta \approx 0$, whereas BASIN consistently shifts $\Delta$ positive. Together, these results identify reasoning basin collapse as a failure mode of inference-time search and show that history-dependent bias provides a simple, training-free mechanism for escaping redundant reasoning under fixed compute. Code is available at https://github.com/GitHubLuCheng/basin
Read the full abstract
Instrumentation and Methods for Astrophysics
Addressing the low-wind effect with the Lyot-based low-order wavefront sensor on SCExAO
The Low-Wind Effect (LWE) is a well-known issue that affects the coronagraphic capabilities of instruments optimized to take direct images of exoplanets. This effect is prominent in segmented and obstructed pupils with spider arms, causing differential pistons due to phase discontinuities and, in some cases, tip-tilt errors under low-wind conditions. LWE-induced low-order errors contribute to coronagraphic leakage and reduce the exoplanet detection sensitivity of direct imaging instruments. Solutions to mitigate the impact of LWE have been developed in both passive mode, such as coating the spider arms with low-emissivity material, and active mode using wavefront sensors (WFS). The Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) instrument at the Subaru Telescope also attempted to mitigate LWE by testing several dedicated WFS. Such techniques, when tested on SCExAO, have proven their ability to measure and correct LWE; however, most remain incompatible with SCExAO's coronagraphic imaging modes. Addressing the need for an efficient solution that can correct LWE in coronagraphic observing modes, we revisited the coronagraphic WFS on SCExAO, known as the Lyot-based Low-order Wavefront Sensor (LLOWFS). LLOWFS utilizes the unused starlight reflected off the Lyot stop and is a well-proven technique that can prevent coronagraphic leaks by sensing low-order errors in a re-imaged focal plane or pupil plane downstream of the focal plane mask. We present the machine-learned LLOWFS's initial on-sky measurement and control of the differential piston aberrations induced by the LWE downstream of a Lyot coronagraph in SCExAO's infrared arm. Preliminary LLOWFS corrections of LWE in the coronagraphic mode are encouraging and would leverage future high-contrast performance of SCExAO, enabling the detection of mature exoplanets at small angular separations.
Read the full abstract
Logic
Modalities in non-classical variations of $\mathsf{S4}$
A classical result in modal logic states that $\mathsf{S4}$ has $14$ modalities, that is, every sequence of negations, boxes, and diamonds is equivalent to one in a set of $14$ such sequences. We study analogous results for the non-classical analogues $\mathsf{CS4}$, $\mathsf{IS4}$, $\mathsf{GS4}$, and $\mathsf{GS4^c}$ of $\mathsf{S4}$. First, we show that, while all these logics have finitely many $\{\Box,\Diamond\}$- and $\{\neg,\Box\}$-modalities, the logic $\mathsf{CS4}$ has infinitely many $\{\neg,\Diamond\}$-modalities. Second, we show that $\mathsf{IS4}$ and $\mathsf{GS4}$ have finitely many $\{\neg,\Diamond\}$-modalities, but they have infinitely many $\{\neg,\Box,\Diamond\}$-modalities. At last, we show that $\mathsf{GS4^c}$ has finitely many $\{\neg,\Box,\Diamond\}$-modalities.
Read the full abstract
Analysis of PDEs
Rigorous Analysis of a Nonlocal Transport–Renewal System for Physiologically Structured Populations
We develop a rigorous analytical framework for a class of physiologically structured population models with two internal state variables, nonlocal ecological feedbacks, dynamic resources, inter-zone transfer, and selective harvesting. The full model is a coupled nonlinear PDE–ODE transport–renewal system with endogenous inflow at the recruitment boundary, a setting in which transport, nonlocal dependence, and boundary renewal interact at the same level. For this full nonautonomous multi-zone system, we prove finite-horizon well-posedness in a positive $L^{1}$-based state space, including global existence on arbitrary bounded time intervals, uniqueness, nonnegativity, and continuous dependence on initial data, environmental forcing, and harvesting effort. We then isolate an autonomous single-zone reduction at extinction and construct a positive compact next-generation operator on the recruit space. In a further nonlinear stationary reduction, we prove that supercriticality of the basic reproduction number $\mathcal R_{0}>1$ yields existence of a nontrivial stationary state under a parametrized compact-operator hypothesis encoding density-dependent renewal feedback. Finally, for a finite-horizon harvest objective over a compact Lipschitz-regular admissible class, we establish existence of an optimal control. The results separate what can be proved for the full climate-explicit system from what can be justified only after autonomous reduction, thereby clarifying the mathematical scope of threshold and control theory for structured populations.
Read the full abstract
Machine Learning
Online Self-Weighted Fine-Tuning
Standard supervised fine-tuning (SFT) assigns the same explicit loss weight to every expert demonstration, regardless of the model's changing competence over training queries. Reinforcement learning (RL) based methods adapt update strength using model-generated rollouts, but often require substantially more sampling and can be unstable on hard tasks. We propose Online Self-Weighted Fine-Tuning (OSW-FT), a simple method that augments SFT with online, trajectory-level weighting. For each query, OSW-FT estimates the model's current success rate using a small number of inference-only rollouts and rescales the standard SFT loss accordingly. The optimization direction remains anchored to the expert trajectory, while the update magnitude adapts online. For binary-verifiable reasoning, we connect this weighting to SFT and RL at the gradient level, inspired by variance-reduction principles. The resulting estimator is unbiased for the exact OSW-FT surrogate update for any finite rollout count, and we analyze convergence with respect to the corresponding surrogate objective. Evaluated across Qwen3 series ranging from 0.6B to 4B on multiple challenging benchmarks (e.g., AIME), OSW-FT consistently improves over SFT on small-to-medium scale models. OSW-FT offers a favorable compute-performance trade-off as a practical approach for fine-tuning small-to-medium LLMs on binary-verifiable reasoning tasks with only 2 online rollouts.
Read the full abstract
Group Theory
Finite permutation groups that act with fixity 4
Motivated by the theory of Riemann surfaces and specifically the significance of Weierstrass points, we prove general structure results about finite groups that have a faithful transitive action with fixity 4. We also explain examples for many different possibilities of such actions.
Read the full abstract
Graphics
Inverse Rig Optimization from Line Drawings
Stylized 3D character animation is largely hand-authored, with animators authoring rig parameters one keyframe at a time to find the best pose. Because stylized work reads chiefly through contour lines, drawing contours in the camera view is the most direct and precise way to express artistic intent. This mismatch between the rig controls and the artist's goal forces a laborious trial-and-error workflow, with animators repeatedly manipulating rig controls against the rendered view to match the desired contour. To address this, we propose a method that recovers rig parameters from screen-space contour strokes, enabling effective keyframing from sketches. Given strokes that redraw the current contour, our method optimizes the high-level rig parameters defined in the DCC tool. The key is to use a pre-trained MLP rig surrogate that provides a differentiable map from rig parameters to mesh vertices, replacing the original black-box rig within the optimization process. We match user-drawn lines to mesh contour lines and backpropagate the resulting screen-space error through the surrogate to update the rig parameters. Our results demonstrate that the method works for diverse characters and practical scenarios.
Read the full abstract
Artificial Intelligence
Agentic Empirical Asset Pricing: Methodological Foundations
Recent advances in LLM agents enable a new paradigm for asset pricing, which we call Agentic Empirical Asset Pricing (AEAP): systems that autonomously conduct the scientific discovery process itself. We define AEAP and identify its core building blocks. Existing evaluation practices backtest only the outputs (factors or trades), not the autonomous discovery system that produced them. We focus on factor discovery, contributing a reference architecture, a rigorous evaluation standard for discovered factors, and a method for out-of-sample backtesting the discovery system. As a concrete instance of that architecture, we evaluate SEADS against five re-implemented baselines on two US equity panels using this standard: no single metric ranks the systems consistently, motivating evaluation on multiple axes at once. A separate rolling re-execution then asks the complementary question of whether the discovery process itself, not one static output, is reliable. We also report negative findings and limitations that surface further evaluation pitfalls for future AEAP systems.
Read the full abstract
Computer Vision
Design and Implementation of a Kalman Filter-Infused Algorithm for Tilt Estimation
Accurate tilt angle estimation is important in many engineering applications, such as robotics, motion tracking, and embedded control systems. However, measurements from low-cost inertial sensors are often degraded by noise and drift. This paper presents a single-axis tilt angle estimation system based on the MPU6050 inertial measurement unit, implemented on an RP2040 microcontroller platform, with sensor fusion achieved through a Kalman filter. The accelerometer provides a direct estimate of tilt angle from gravity but is sensitive to noise and short-term fluctuations. The gyroscope provides smooth angular rate measurements, but integration over time introduces drift. To overcome these limitations, a Kalman filter is used to combine measurements from both sensors, leveraging the long-term stability of the accelerometer and the short-term smoothness of the gyroscope. Both simulation and hardware experiments are performed. In simulation, sensor noise and drift are modeled to evaluate the filter performance under control conditions. In the hardware implementation, real-time MPU6050 data is acquired and processed by the RP2040 platform, and the estimated tilt angle is compared with accelerometer-only and gyroscope-only outputs. The results show that the proposed method effectively reduces noise measurements and suppresses long-term drift while preserving good dynamic response. Overall, the system provides more stable and accurate tilt estimation than either sensor alone, demonstrating a practical and accessible approach for Kalman filter based sensor fusion in embedded application. This manuscript is a preprint version of the work. Keywords: Kalman Filter, Accelerometer, Gyroscope, Noise Reduction, Angle Tracking
Read the full abstract
Mesoscale and Nanoscale Physics
Classifying coherent peaks in nanoelectronic devices by the presence or absence of spin exchange
Coherent peaks appearing in the differential conductance of quantum-dot single-electron transistors (QDSETs) and quantum point contact (QPC) devices are classified into two categories according to the scaling function onto which the temperature-scaled differential-conductance maxima collapse and the underlying spin dynamics. The zero-bias peaks (ZBPs) observed in QPCs and in the triplet state of the even-particle sector of QDSETs belong to the same category, whereas the ZBP in the odd-particle sector of a QDSET belongs to a different category together with all finite-bias coherent peaks observed in QPCs and in the even-particle sector of QDSETs. The spin dynamics of the former category involve spin exchange, a hallmark of Kondo dynamics, whereas those of the latter category involve only cotunneling of an up–down spin pair. Furthermore, for the former type of ZBP, the scaling temperature coincides with one-half of the full width at half maximum (FWHM), which corresponds to the Kondo temperature. In contrast, for the latter type, the scaling temperature does not coincide with the (1/2)FWHM-derived energy scale. To support these findings, the gate-voltage-dependent differential-conductance line shapes measured in the odd-particle sector of a QDSET are theoretically reproduced. The results demonstrate that the observed ZBP is a merging of two coherent side peaks generated solely by the cotunneling of up–down spin pairs.
Read the full abstract
Artificial Intelligence
SOVER: Formal Certification of Optimization Reformulations via LLM-Assisted SMT Verification
Large Language Models (LLMs) have shown remarkable promise in translating and reformulating complex mathematical optimization problems across modeling languages. However, validating such transformations through empirical solver executions alone is unreliable, as solver outcomes may be affected by local minima, structural timeouts, numerical artifacts, and subtle semantic divergence between formulations. We introduce SOVER, an LLM-assisted SMT framework that separates semantic mapping from formal certification: Z3 checks domain cross-feasibility and global objective-order preservation for mixed-integer linear formulations, while dReal provides tolerance-aware feasibility/range and $\epsilon$-argmin checks for continuous nonlinear formulations. We also introduce NLEquiv-150, a public benchmark of 100 equivalent and 50 deliberately hard non-equivalent nonlinear reformulation pairs. With LLM-extracted mappings, SOVER classifies 149/150 pairs (99.33%) correctly, including all 50 hard negatives; the sole error is an incomplete mapping extraction.
Read the full abstract
Sound
Heard but Not Heeded: Paralinguistic Information Encoding and Loss in Audio-Language Models
Audio language models are designed to understand speech, yet it remains unclear whether they capture how something is said beyond what is said. We present a mechanistic analysis of paralinguistic information in four open source models, Whisper-large-v2, Qwen2-Audio-7B Instruct, Qwen2.5-Omni-7B, and Chroma-4B, using the Expresso dataset with controlled speaking styles. We combine centered kernel alignment, linear probing with leave one speaker out evaluation, open ended tone prediction, and a content prosody leakage metric to trace how style information moves from the audio encoder to the final output. All models strongly encode speaking style in the late encoder, that is, the top third of the audio encoder's layers, but this information is consistently degraded before reaching the output. The projector reshapes representation geometry without removing information, while decoders differ in how much style they preserve depending on architecture and training objective. At the output level, models fall into two behaviors. Some are content driven, where predictions depend mainly on text. Others are acoustic driven, where predictions vary with speaking style. The leakage metric quantifies this difference, and qualitative results confirm it. Overall, we identify a gap between what models encode and what they use, highlighting a key limitation in current audio language models.
Read the full abstract
Representation Theory
Scary Wheels (and Super Shrubs)
We prove the shuffle realization for quantum affine superalgebras of types $B,C,D$ in the loop realization, in that we construct an isomorphism between each of these algebras and a suitably defined double shuffle algebra. We explicitly describe the latter shuffle algebra using certain vanishing conditions that generalize the Feigin-Odesskii wheel conditions; a novel feature is the appearance of a so-called scary wheel, which is a particular order $2$ vanishing condition involving $7$ variables. Along the way, we fully develop the theory of shrubs in super types $A$ (affine and toroidal), which are important combinatorial tools in the study of the shuffle algebras associated to toric Calabi-Yau threefolds. Our techniques also allow us to define quantum toroidal superalgebras of types $B,C,D$. More importantly, we provide a general framework to formulate and prove shuffle realizations in the wide generality of quivers with parameters.
Read the full abstract
Quantitative Methods
Efficiently classifying shocks in complex systems requires dormant reporters
Many natural and engineered systems are large complex networks of interacting components, and external perturbations drive them along different dynamical paths. Identifying which perturbation occurred matters for diagnosis, control, and prediction. Yet often times only a few components can be jointly monitored. Which components should be monitored? Experimental practice usually favors placing reporters at the most sensitive sites, where perturbations produce the largest effects. Using a simple dynamical model for complex systems with heterogeneous connectivity, we ask how sparse reporter panels should be chosen to classify shocks from partial trajectories when repeated trials only approximately reproduce an ideal initial condition. Once that reproduction is imperfect, sensitivity ranked panels fall far short of optimal, and the shortfall grows with the noise. We find that the best panels mix two kinds of reporters. A promiscuous reporter responds to most shocks, so it separates them mainly by degree, and degree fluctuates from trial to trial. A dormant reporter responds to only a few shocks but strongly, and its answers do not scatter as much between trials. As noise grows, the cost of losing a dormant reporter rises to meet the cost of losing a promiscuous one. Panels of either kind alone classify worse than the mixture, and no property of the members collected individually explains the ordering. Most of all, we find that only a minuscule number of reporters are needed on a panel to accurately identify which shock hit the system. We implement an efficient algorithm to assemble such a panel. Together these results provide a low cost practical design principle for monitoring large complex dynamical systems.
Read the full abstract
Materials Science
Phase Switchable Photocatalytic Water Splitting via a Paraelectric-Ferroelectric Transition in Zr2Ge2S6 Monolayer: A Comprehensive Theoretical Insights
Photocatalytic water splitting (PWS) is a promising technology for addressing the global energy crisis and producing renewable and clean hydrogen fuel. Although numerous 2D materials have recently been proposed as potential photocatalysts, effective strategies for regulating photocatalytic reactions and improving energy conversion efficiency remain limited due to performance regulation challenges. Here, using first-principles calculations, we demonstrate that the photocatalytic activity and energy conversion efficiency of a Zr2Ge2S6 monolayer can be effectively tuned through a paraelectric-ferroelectric phase transition. The Zr2Ge2S6 monolayer exhibits excellent structural stability, favorable mechanical properties, a suitable band gap, optimal band edge positions, and broad-spectrum light absorption. Moreover, the Zr2Ge2S6 monolayer exhibits a higher oxidation potential and a stronger driving force for photogenerated holes to promote oxygen evolution reaction (OER) in the ferroelectric phase. In contrast, the paraelectric phase provides photogenerated electrons with a greater reduction potential and driving force for hydrogen evolution reaction (HER). The solar-to-hydrogen conversion efficiency is also strongly influenced by the phase transition, increasing from 7.71% in the paraelectric phase to 15.31% in the ferroelectric phase because of the improved carrier utilization. Our theoretical investigation not only highlights the crucial role of ferroelectric polarization in photocatalytic water splitting but also provides an effective strategy for tuning the photocatalytic properties of 2D ferroelectric materials through ferroelectric switching.
Read the full abstract
Algebraic Geometry
An integral Hyodo–Kato isomorphism
Let $\mathscr{O}_K$ be a mixed characteristic complete DVR with perfect residue field $k$, and let $\mathfrak{X}$ be a proper formal scheme over $\mathscr{O}_K$ with semistable reduction. Answering a question of Fontaine and Jannsen, a classical theorem of Hyodo and Kato gives a rational identification between the log crystalline cohomology of the special fiber $\mathfrak{X}_0$ over $W(k)^0$ with the de Rham cohomology of the generic fiber $\mathfrak{X}_K$. In this note, we use a log variant of the saturated de Rham–Witt complex of Bhatt–Lurie–Mathew to prove that such a comparison holds integrally.
Read the full abstract
Mesoscale and Nanoscale Physics
Process-Technology Co-optimization for 2D-FETs
We present the first experimental machine learning (ML)-enabled Process-Technology Co-Optimization (PTCO) framework for optimizing 2D transition metal dichalcogenide (TMD) FET fabrication directly from statistically meaningful experimental data rather than pure simulation data. We first introduce a transition voltage metric, VTrans, to quantify the gate voltage required for off-to-on switching and reveal its direct correlation with subthreshold swing (SS), highlighting an overlooked switching characteristic that governs both off-state and on-state performance. By integrating automated metric extraction, multi-objective recipe ranking, and predictive modeling, our framework uncovers hidden process-performance correlations and predicts the performance of unexplored fabrication recipes from limited experimental data. Experimental validation shows close agreement with ML predictions, thus demonstrating the framework's ability to efficiently guide gate stack optimization through iterative experimental feedback.
Read the full abstract
Statistical Mechanics
BKT-like Correlation Scaling and Twist Responses in a One-Dimensional Fractional $U(1)$ Ginzburg–Landau Model
We study a one-dimensional fractional $U(1)$ Ginzburg–Landau model whose quadratic part has Fourier multiplier $|k|^\sigma$, focusing on the marginal case $\sigma=1$. This dispersion yields logarithmic spin-wave fluctuations, suggesting BKT-like behavior despite the one-dimensional setting. We sample the equilibrium Gibbs measure using stochastic Gross–Pitaevskii dynamics and analyze correlation functions, dimensionless ratios, effective exponents, and twist responses. The correlation function shows a low-temperature algebraic branch with a temperature-dependent exponent, while the high-temperature regime exhibits a nonlocal-kernel-induced tail consistent with $C(r)\sim r^{-2}$. The correlation and Binder ratios are nearly size independent at low temperature and collapse with the BKT-type variable $(T-T_{\rm BKT})(\log L)^2$; finite-size effects set in around $T\simeq0.35\text{--}0.4$, consistent with $T_{\rm BKT}\simeq0.35$. Unlike the two-dimensional XY model, twist responses do not yield a finite helicity modulus: the ordinary linear-response quantity grows with system size, whereas the cusp twist response scales as $L^{-\eta(T)}$, like the squared zero-mode order parameter. Thus, the transition is BKT-like in correlation scaling, but lacks a universal helicity-modulus jump.
Read the full abstract
Combinatorics
Improved bounds for the lazy cops and robbers on generalized hypercubes
In Lazy Cops and Robbers, at most one cop moves on each cop turn. We study the lazy cop number of the generalized hypercube $Q(n,m)$, whose vertex set is ${\{0,1,\ldots,m\}}^n$. For each fixed integer $m\geq2$, we prove the asymptotic upper bound $$c_{\mathrm{L}}(Q(n,m))=O\!\left(\frac{{(m+1)}^n}{n^{3/2}}\right).$$ This result improves the upper bound of Sim, Tan, and Wong by a factor of $\log n$. The proof combines a moving dominating-set argument with an explicit dominating-set construction inside the support classes of each level. As a separate domination result, we show that, for fixed integers $m\geq2$ and $d\geq1$, the Hamming graph $K_m^{\square k}$ has a distance-$d$ dominating set of asymptotic size $O(m^k/k^d)$. This order is optimal up to a constant factor.
Read the full abstract
Analysis of PDEs
On the Aleksandrov–Bakelman–Pucci estimates for the weighted $1$-Laplacian
We establish Aleksandrov–Bakelman–Pucci-type estimates for viscosity subsolutions of Poisson equations associated with the weighted $1$-Laplacian and, more generally, with fully nonlinear operators acting on the Hessian in directions orthogonal to the gradient. A key feature of the proof is that the quasiconcave envelope plays the role of the concave envelope in the classical ABP estimate.
Read the full abstract
Artificial Intelligence
A Closed-Loop Evaluation of Capability Loss and Recovery in Compressed Driving Policies
Many automobile and mobility companies deploy learned driving policies on embedded computers with limited memory and power. Pruning, knowledge distillation, and quantization are the standard methods to reduce the size and the inference cost of these policies. However, these methods are commonly assessed by aggregate numerical scores, and such scores may not reflect the ability of the policy to drive safely when interacting with other road users. In this study, we propose a stage-wise closed-loop evaluation approach to follow a driving policy through a compression pipeline. We formulate the driving task as a partially observable Markov decision process (POMDP) and train a belief-state policy with proximal policy optimization (PPO) in Gym-Duckietown. We then extract the actor, compress it one stage at a time, and evaluate it on five driving curricula. We show that structured pruning is the stage at which the driving capability is first lost. Meanwhile, distillation improves the pruned actor, but the improvement is limited by its rehearsal data. Integer quantization of the improved actor loses some of the curricula that require the vehicle to stop and then resume. Interestingly, the same procedure on the unpruned actor preserves all five curricula. Our study thus provides an empirical analysis aiming to answer the currently active discussions on how to accept a compressed driving policy, so as to achieve a safe and statistically reliable deployment of automated driving functions.
Read the full abstract
Statistics Theory
Extending Subsampling to Sequential Stopping
Fixed-width sequential stopping rules terminate a stochastic simulation once an estimated confidence interval reaches a prescribed width. Classical fixed-width theory typically relies on a strongly consistent estimator of the asymptotic variance. This makes the normalized stopping time asymptotically deterministic, allowing fixed-sample-size limit theory to be transferred to the estimator at termination. This mechanism can fail when simulation output has infinite variance or long-range dependence. Although self-normalization and subsampling can yield asymptotically valid confidence intervals at a fixed sample size, the scaling process and the stopping time may retain nondegenerate randomness, so fixed-sample-size quantiles need not provide valid coverage at termination. In this paper, we develop a unified framework based on a joint functional limit theorem for the estimation process and a scaling process. We characterize the asymptotic behavior of both the stopping time and the self-normalized estimator evaluated at termination, thereby obtaining asymptotically valid sequential confidence intervals in classical finite-variance and infinite-variance settings. Moreover, we introduce a sequential subsampling procedure that consistently estimates the distribution relevant at the stopping time without directly estimating nuisance parameters in the limit distribution. The framework is verified for heavy-tailed moving-average processes, stochastic approximation, and an M/G/1 queue with heavy-tailed service times.
Read the full abstract
Human-Computer Interaction
GazeTune: Facilitating Precise Gaze-Driven Interactions with Cascaded Touch Input
Eye gaze has become an essential input for spatial computing, but its coarse targeting and saccadic nature limit precision and complicate continuous interactions such as dragging, especially under user motion. Gaze+pinch has also become standard in XR for its convenience, yet mid-air gestures remain imprecise, fatiguing, and socially unacceptable. These limitations underscore the need for an approach that preserves the speed of gaze while enabling stable, fine control. We present GazeTune, a cascaded multimodal interaction technique combining gaze and touch to refine gaze-based selection and manipulation. Touch serves as a refinement channel within gaze pointing, allowing precise cursor and target control. Our work investigates how gaze-and-touch enhances dragging and mitigates Motion-Induced instability. In a study (N=20), we compared GazeTune against gaze-only and gaze-pinch methods in 2D dragging. Results show that GazeTune achieves significantly lower error with comparable execution time, validating its effectiveness and balanced trade-off between time and accuracy.
Read the full abstract
Information Theory
MaskCode: Mask Transformer for Feedback-Assisted Coding With Linear Block Codes
Feedback-based coding schemes have demonstrated substantial performance gains over today's open-loop coding schemes. Unfortunately, these gains are usually achieved in idealized settings with perfect feedback. Over the last few years, machine learning-based schemes have been shown to be promising solutions for implementing feedback-based codes, particularly when combined with short-block-length open-loop error correcting codes (ECCs) in a concatenated coding structure. However, existing ML-based feedback schemes remain agnostic to the outer code's structure, potentially misallocating feedback resources on error patterns already correctable by the outer ECC. To address this, we propose MaskCode, a Transformer-based inner feedback code for concatenated coding systems, which explicitly incorporates structural knowledge of the outer linear block code into the inner feedback encoder design via two synergistic mechanisms: 1) a soft syndrome-based input that informs the encoder about potential parity constraint violations, and 2) a code-aware attention mask derived from the Tanner graph. We further show that end-to-end training with a differentiable belief propagation (BP) decoder offers no additional gain, as MaskCode's structure-aware design already internalizes the structural knowledge of the outer code; in fact, backpropagation through the iterative BP decoder introduces gradient explosion, which degrades rather than improves performance. Extensive evaluations on BCH and LDPC outer codes demonstrate that MaskCode consistently outperforms all baselines, achieving up to 1.5 dB SNR gain.
Read the full abstract
Artificial Intelligence
ChatDev 2.0: A No-Code Multi-Agent Platform for Developing Everything
Large language model (LLM)-based multi-agent systems (MAS) have shown strong potential for solving complex tasks, yet their development forces a tradeoff: code frameworks are expressive but engineering-intensive, while no-code builders simplify authoring but constrain agent interactions to author-defined workflows. We present ChatDev 2.0: DevAll (hereafter DevAll), a no-code platform for building, executing, and inspecting heterogeneous MAS that delivers both high expressiveness and ease of use. In terms of expressiveness, DevAll pairs a declarative executable graph abstraction with a cycle-aware execution engine, so that heterogeneous agents and dynamic and cyclic interactions can be represented and executed within a single framework. For ease of use, an integrated visual interface lets users author, run, monitor, and inspect MAS, including human-in-the-loop steps, entirely without writing code. Experiments demonstrate that DevAll reproduces state-of-the-art MAS across three representative tasks at competitive performance and without task-specific orchestration code, highlighting its effectiveness as a general-purpose platform for LLM-based MAS. DevAll is available at https://github.com/OpenBMB/ChatDev.
Read the full abstract
Computer Vision
Efficient and Robust Absolute Pose Estimation via Gravity-Prior-Driven Transformation Decoupling and Pose Refinement
Estimation of the absolute pose of an object is an essential task for various robotic applications. Recently, incorporating gravity direction as prior information has emerged as a popular approach to simplify absolute pose estimation. However, developing a robust and efficient algorithm to solve this challenging problem remains a difficult question due to large amounts of mismatches. In addition, obtaining an accurate pose solution from selected inlier correspondences with gravity prior is still a research gap. In this paper, we propose a novel transformation strategy that exploits geometric relations derived from the gravity prior. Through transformation decoupling, the original 6 degrees of freedom (DoF) absolute pose estimation problem is simplified into a 4-DoFs problem: 1-DoF for the rotation angle and 3-DoFs for translation, significantly improving the efficiency. For the 1-DoF rotation angle, we apply a one-dimensional global voting algorithm for optimal estimation. Once the optimal rotation is obtained, the mismatched correspondences are preliminarily filtered, and translation estimation, a linear problem, can be easily solved. Furthermore, to obtain accurate pose results, we introduce a novel pose refinement algorithm to enhance the accuracy of both rotation and translation. Extensive experiments on synthetic data and three publicly available real-world datasets (TUM RGB-D, ETH3D, and RobotCar) demonstrate that the proposed method achieves stronger performance compared to existing state-of-the-art (SOTA) approaches. To further validate our method, we integrated it into ORB-SLAM2. The results on the KITTI dataset show it effectively reduces drift and improves trajectory alignment during relocalization. The source code will be released upon acceptance.
Read the full abstract
Computer Vision
EarthLD: Towards Unified Open-World Landslide Understanding via Vision-Language Guided Diffusion Models
Landslides are widespread geological hazards, yet their automated detection and mapping in remote sensing imagery remain challenging because of their irregular morphology, ambiguous spectral signatures, and substantial domain shifts across imaging platforms. To overcome these challenges, we propose EarthLD, a vision-language-guided diffusion framework for open-world landslide understanding, enabling unified landslide recognition, mapping, and trigger interpretation. At its core, EarthLD formulates landslide understanding as a diffusion process that progressively infers the presence, spatial extent, and pixel-level boundaries of landslides from noisy latent representations. This probabilistic formulation enables the model to jointly perform image-level landslide recognition and mapping while characterizing predictive uncertainty. By integrating visual observations with contextual knowledge in the denoising process, EarthLD distinguishes diverse landslides from backgrounds, produces confidence-aware predictions for suspected regions, and maps landslide ranges. We additionally construct a global-scale open-world landslide benchmark by systematically harmonizing multiple publicly available remote sensing data collected by diverse institutions. Extensive experiments across regions, sensors, and triggering events demonstrate that EarthLD consistently outperforms existing landslide detection methods, highlighting its potential as a unified and robust solution for global geological-hazard monitoring and emergency response.